My stm32 powered pi detector

[Mdtoday]

Latrobe valley Vic mate..
Your spot on with the pleasing everyone comment but a lot of good has come of it yes the design is changing but so many valid points I just did not think of and am happy to fix, the general design still remains.
Costs are down, size is down.
I am proceeding with smd but will port to a modular design also.
I have not thought about selling boards but it's a defendant possibility I will be ordering stencils with my PCBs so it would make sense to send boards out with components attached.
There will also be 3d printer friendly enclosures to come.
Thanks for the front end leads..

My old stomping ground, Thompson river, Tangell Bren, Bairnsdale, some nice little goldfields through there and like you say heavy mineralization. I'm in SA.
Yes, there are a lot of helpful people on this site with a wealth of experience and it's good to hear you have your general design in tact.
As Waltr has mentioned, the front end is where board design is critical, star ground, low noise amp, input banking, damping but just as important is your TX supply.
The ADC I use is an ADS8329 and if you are looking fir great 16 bit Dacs have a look at DAC8568 from TI very handy device.
I have a reasonable amount of gain in front of the ADC in my design and if I did it again, I would use a minimum of 18 - 19 bits. So basically a 24 bit ADC.

On on selling loaded boards, I think that's a good move as long as you can handle " mine doesn't work" questions.
3d printing, now your talking!


I am sure you will come up with a great design.

btw, just ordered a PWD13F60 full bridge development board from Element14, I'm quite interested to see how this will perform

Cheers

MDtoday

[gtdavid]

Have you received your dev board yet?

[Mdtoday]

No was hoping to have received today but no show..supplier said 6-7days so must be close..

cheers

Mdtoday

[Mdtoday]

Hi gtdavid,


Received PWD13F60 evaluation board this morning, looks Ok, bit expensive at $57 delivered but chips themselves are around $7 so not too bad.

I'll try set up some tests when I get some time and post when done.

[Mdtoday]

here are some better closeup photos




.1" headers will make connecting easy.

cheers

Mdtoday

[gtdavid]

That is a bit "OK" isn't it, expect at least a clean board from ST.

[Mdtoday]

Yes, pins were soldered in at odd heights too but hopefully it will work as expected.

If it does, I would be inclined to make a smaller version of the board but have to check out heat dissipation of this board in actual operation with a coil.

[Mechanic]

Very interesting, I like where this is going!
Don't forget to scatter a few fiducials about the place in case you need some pick and place done. Does Kikad export pick and place data?
And as otheers have said, front end is very important. Low noise, high speed, fast recovery and as much dynamic range as your chosen adc(more bits will be better) can handle. Variable analog gain in the reveiver would be good too to adjust the gain to the ground signal.

Cheers Mick

[Mdtoday]

"Variable analog gain in the receiver would be good too to adjust the gain to the ground signal"

Great points Mick.

Yes, that is something I use on my design, along with an very stable but adjustable ADC reference voltage. Both digitally controlled.
I would include at least adjustable gain if possible.

Not sure on the Kicad exports...I use Altium but it appears to be a .pos file (CSV) which needs converting, not sure?

https://forum.kicad.info/t/xy-parts-...-and-place/368

cheers

MDtoday

[gtdavid]

Variable gain on rx is currently on both first and second amp stages. Currently working on the tx.
Sorry about the delay on the github page I will get there I have redone the rx about 5 times now once everything is hashed out I will put something up

[gtdavid]

Pick and place can be done just have to remember how

[aa1ww]

Finding this thread is a neat coincidence. I just got an STM32H743ZI Nucleo board from Mouser ($23 + $7 S/H). I'm looking at WM8731 or CS4272 + ucontroller for an SDR project. Like almost every 32bit ucontroller vendor, ST provides a free Eclipse-like IDE and a HAL & CMSIS code generator (called STM32CubeMX). These two perform nearly the same functions as MPLAB-X and Harmony for the PIC32 families. The grunt work I'm thinking about is in the DSP (or other tight computational code) and, while 400MHz sounds great (heck, it is great) one needs to delve into the ARM Cortex (r) M architecture docs to see how to keep code in TCM (tightly coupled memory) or DCM or whatever to achieve the highest performance. Truthfully, I haven't finished analyzing FIR MAC performance between dsPIC33E, PIC32MZ-ECH, and STM32H7. By the way, NXP is offering a 600MHz Cortex (r) M7 by way of their acquisition of FreeScale and its i.MX RT1050 "Crossover". The dev board costs (gulp) $84USD @ https://goo.gl/JT4x2R. The same caveats about wait states and MACs applies, of course.

re ADCs: bear in mind that another approach to resolution is found in rate conversion (e.g. decimation) and one might realize advantage in that approach. DMA and circular buffers, once one does all the low-level setup work, really streamline the data flow for signal processing.

Anyway, kudos to you folks for brainstorming a pretty intriguing framing of this thing.

Now, back to trying to catch up on a couple of decades of ... what do you call these things? ... metal detectors, that's it. Great forum.

Coop, aa1ww

[Mdtoday]

Hi aa1ww,

There are definitely some good dev boards out there for sure, its just a matter of picking the right flavour.

On the codecs, the thing to watch out for is the use of internal high pass filters which, in some devices (not all), can't be switched out and/or are too high in pass band for reasonable use in a PI.

I have tried a number of devices and not been all that happy with the performance in my particular design and ended up reverting back to a single channel high resolution ADC.

Not saying it can't be done, it has but I don't know what codecs are being used.

With out pulling this thread sideways too much, it would be interesting to hear from other members that have had success with codecs, which devices were used.

cheers

MDtoday

[Mdtoday]

Variable gain on rx is currently on both first and second amp stages. Currently working on the tx.
Sorry about the delay on the github page I will get there I have redone the rx about 5 times now once everything is hashed out I will put something up

I'm planning to fire up the bridge driver over the weekend, I have been snowed under with work lately, so haven't had much free time.
Hopefully I can post some positive results.

BTW meant to ask, are you planning to use LI cells as the power source?

It would be nice to use a boost reg with a couple of 18650, that will help reduce weight.

cheers

MDtoday

[gtdavid]

I use LI 18650 cells in everything I do, then couple with a charge/balance controller.
There is boost circuit for the tx already but I would not go lower than a 4 cell +/- 16v, but how many in parallel would be up to you.
I have been doing some LTspice sims trying to cross over Thinkerer's IB research with mixed results, the problem is the bucking coil and bipolar operation.
On the ADC side, I think I will go with the AK4558EN as there is plenty of documentation out there for troubleshooting.